A Study On P300and SSVEP-based High-performance Brain-computer Interface And Its Application

Brain-computer interfaces (BCIs) provide non-muscular communication and controlby directly translating brain activities recorded from the scalp into computer control sig-nals and thus enable users with motor disabilities to convey their intent to the externalworld. An important issue in BCI research is high-performance BCIs, which enables con-siderably enhanced interfacing between the user and the machine. Potential applicationsinclude BCI controlled wheelchair, consciousness detection in patients with disorders ofconsciousness, etc.First, a comparison of two existing P300spellers (Single Display (SD) speller andRegion-based (RB) speller) was presented. In our experiment, higher online accuracywas obtained for RB speller than for SD speller. Further data analysis results, includingP300detection, P300waveform and Fisher ratio, demonstrated that P300potential wasenhanced in RB speller. This enhancement, which led to better performance of BCIspeller, might be due to the increased distance and the decreased interference betweenneighbor buttons. Our study also suggests that when we design a stimuli display paradigmfor P300detection, several factors such as the probability of oddball occurrence and thedistance between adjacent buttons should be considered simultaneously.Second, a pseudo-key-based approach is proposed to discriminate the control andidle states for an asynchronous steady state visual evoked potential (SSVEP)-based brainswitch. Unlike the existing brain switches without pseudo-keys that use a thresholdingmethod to distinguish between control and idle states, the detection in the proposed ap-proach is based on a threshold condition and a comparison condition. Specifically, theintroduced comparison between the power of the target button and the pseudo-keys cansignificantly improve the discrimination of the control and idle states. In our experiments,we compared the pseudo-key-based brain switch with the one without pseudo-keys. Dataanalysis indicates that the pseudo-key-based brain switch produced more accurate per-formance (accuracy86.67%, true positive rate7.49event/min, false positive rate0.98event/min). With this study, we won the second place of the Switches control event inthe1st China BCI Competition (2010).Next, a hybrid asynchronous BCI combining P300and SSVEP was presented, inwhich the control state and target button are determined by both P300and SSVEP de- tections. Our experimental results showed that both P300and SSVEP can be elicitedsimultaneously and that the performance for detecting the control/idle state can be im-proved using our hybrid BCI. In BCI wheelchair control, it is an important and difficulttask to produce a “stop”command as accurately and quickly as possible. As an appli-cation, we used our hybrid BCI system for the “go/stop”control of a real wheelchair,and demonstrated our system’s effectiveness (response time5.28s, false activation rate0.52event/min) in an online wheelchair control experiment.Based on the hybrid BCI, a variant of our previous system was successfully usedto detect awareness in patients with DOC. To our knowledge, this is the first attemptto test a hybrid BCI in this population. The BCI system determined which photo wasfocused on by the patient through both P300and SSVEP detections. Eight patientsattended our experiment,3of which were able to follow the commands using our hybridBCI (classification accuracy70%-78%). Moreover, both P300and SSVEP were elicitedsimultaneously for the3patients who could perform command following. This impliedthat the3patients had residual cognitive function and conscious awareness, which wasdetected by our BCI system.